Critical analysis of Unsupervised Learning methods of Dimensional modeling

Critical analysis of Unsupervised Learning methods of Dimensional modeling

Unsupervised learning method for dimensional modeling:

Unsupervised learning method will be selected for such situation where you don’t of have the information about data by which you can classify it as FACTS and DIMENSIONS. Even in some cases you can not put any information indirectly and let the modeling go as it goes by default. This method works with one-way clustering and two-way clustering. In one-way it uses all the attributes and in two-way or bi-clustering or local view of your data set.

Shortcomings:

1- Query Performance because of data redundancy
2- Useless matching, pairing in shape of similarity and dissimilarity matrix
3- Will not represent true picture of data not because of data quality but because of unidentified attributes e.g. quantitative attributes are selected as descriptive as well or vice versa.